The point break effect: an examination of surf, crime, and transitory opportunities

Abstract

Opportunity theories of crime suggest that crime is highly specific and concentrated in time and space. Using these theories as a framework, this paper seeks to examine the transitory nature of crime. This hypothesis was tested using data from a coastal city in California to examine the relationship between surf conditions (measured at five daily time points) and number of crime incidents (n = 16,075). Crime totals were aggregated at the street segment level (n = 4551) for each day in 2011. These data were modeled using a series of panel negative binomial models, clustered by census block group. The findings suggest that surf conditions had an effect on the likelihood of crime incidents, but these effects were time specific. Favorable surf conditions were associated with increases in crime only between 2:30 pm and 5:29 pm. Additionally, locations closer to surf spots were associated with more crime, relative to locations farther away. Closer examination of micro-geographies aids in understanding how systematic shifts in routine activities affect the frequency and location of crime, and allows crime prevention to be more specialized and efficient. Adding to the extant understanding of hot times and opportunity structures will enable more effective allocation of resources and predictive policing efforts.

Keywords:

• policing
• hot times
• opportunity
• predictive policing
• hot spots
• routine activities

Acknowledgements

The authors would like to acknowledge and thank Michael White and Hank Fradella for their helpful comments on earlier drafts of this manuscript, and to Joseph Hoploc and Aili Malm for conversations that helped shape our ideas. We would also like to extend thanks to Alyssa Chamberlain, Gary Sweeten, Danielle Wallace, and Courtney Riggs for consulting with us on the methodology. Additional thanks are extended to the three anonymous reviewers for their helpful insights. We are grateful for the data provided by the Ventura Police Department and the information we gained in discussions with the California Department of Parks and Recreation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Routine activities theory specifically draws upon Hawley’s (1950) space-time concepts of rhythm (i.e. the regular periodicity with which events occur), tempo (i.e. the number of events per unit of time), and timing (i.e. the coordination of different interdependent activities). In short, Hawley’s (1950) space-time concepts frame ‘the coordination of an offender’s rhythms with those of a victim’ (Cohen & Felson, 1979, p. 590). Building on this notion, Felson (2006) suggests that one avenue in which offenders may align their rhythm – or routine activities – with suitable targets is through overlapping activity spaces (i.e. where people live, work, shop, or seek entertainment). These events and activities, however, may not be as routine as criminologists traditionally suggest; sporadic events can be systematic (Felson, 1987).

2. It should be noted that Internet use facilitates the organization of collective action and various behaviors in a number of ways, including protests, riots, flash mobs, pedophilia, gang activity, and cybercrimes (Bennett & Segerberg, 2011; Brejzek, 2010; Brenner, 2002; Deirmenjian, 2002; Densley, 2013; Glenny, 2011; Mann & Sutton, 1998; Pyrooz, Decker, & Moule, 2015; Segerberg & Bennett, 2011; Sela-Shayovitz, 2012; Wall, 2007; Wasik, 2009; Wolak, Finkelhor, & Mitchell, 2004). Such participation and spatial organization is considered a physicalization of viral culture (Brejzek, 2010; Wasik, 2009).

3. The police incident reports used in this study include a variety of crime types (e.g. violent crime, property crime, driving under the influence, disturbing the peace, etc.). The Ventura Police Department requested that sexual assault data be removed from the data-set. A total of 54 cases were removed.

4. ‘Even the most active offenders do not commit crimes around the clock’ (Bernasco, Ruiter, Bruinsma, Pauwels, & Weerman, 2013, p. 896), and human behavior is constrained by physiological necessities (Hägerstrand, 1970; Pred, 1973).

5. Of all crime incidents occurring on street segments (n = 14,299), 32.1% of these cases were excluded from the analysis because they did not fall within a catchment time period (crime occurred between 8:30 pm and 5:29 am). The distribution of the crime incidents by catchment time that are included in the analysis is as follows: 9.8% between 5:30 am and 8:29 am, 18.6% between 8:30 am and 11:29 am, 23.0% between 11:30 am and 2:29 pm, 25.7% between 2:30 pm and 5:29 pm, and 22.9% between 5:30 pm and 8:29 pm.

6. The time span that crime may occur on good surfing days, near the beach, and surrounding areas, for example, could be expected to be approximately from 6 am to 6 pm.

7. The following equation was used to compute the Euclidean distance in decimal degrees: .

8. The issue of spatial autocorrelation arises when nearby values tend to be more similar than distant values (Tobler, 1970). A spatial lag term was created in GeoDa to account for significant, positive spatial autocorrelation in the dependent variable (Moran’s I = .03; p < .001) using a 2.5-mile Euclidean distance decay function (Rengert, Piquero, & Jones, 1999). However, the spatial lag term was excluded in the final model because it is highly correlated with the Euclidean variable (r = −.947; rho = −.965). Accordingly, spatial autocorrelation is adequately controlled for by the inclusion of the Euclidean distance variable.

9. It is assumed that standard error of segments is correlated within block groups but independent across block groups (see also Di, Murdoch, & Ma, 2009).

10. The clustering of standard errors was selected over other methodological approaches because street segment level data are not provided by the US Census (e.g. measures are not available to control for differences between segments such as poverty levels). Other statistical approaches to modeling the data would preclude the modeling of key variables of interest. For instance, fixed effects regression model would control for unmeasured differences between street segments, but this technique does not allow for the inclusion of time invariant covariates (such as Euclidean distance from surf location).

11. In a 2013 article, Ray Bergman, a surfer and journalist, recalls an interaction he had with a man breaking into his car while he was surfing. The offender stated that there are several indicators of lack of guardianship that he looks for before offending: surf stickers (which indicate who is surfing and will be away for an extended period of time), secluded parking spots, unlocked doors/windows, hiding a key on a vehicle where someone can easily find it, owning an older car that is easier to break into, and so on. The Bergman (2013) interview does provide insight into an offender’s motivation, but a larger sample would be able to ascertain if crime at surfing locations is strictly opportunistic, premeditated, or a combination of these.